sign in sign up
<<
Home , Ulcerative colitis

Dig Dis Sci (2015) 60:492–501 DOI 10.1007/s10620-014-3373-2

ORIGINAL ARTICLE

The Risk of Colorectal in Patients with Ulcerative

Tobias M. Nowacki • Markus Bru¨ckner • Maria Eveslage • Phil Tepasse • Friederike Pott • Nils H. Thoennissen • Karin Hengst • Matthias Ross • Dominik Bettenworth

Received: 11 April 2014 / Accepted: 24 September 2014 / Published online: 4 October 2014 Ó Springer Science+Business Media New York 2014

Abstract associated with the risk of CAC (P \ 0.0014); disease Background and Aim Ulcerative colitis increases the risk duration between 9 and 15 years: OR of 2.5 (95 % CI of developing and colitis-associated cancer 0.2–41.1), more than 15 years: OR of 21.4 (95 % CI (CAC). The purpose of this study was to determine the risk 2.6–173.6). The risk of developing dysplasia (low-grade factors as well as protective measures for disease burden, intraepithelial neoplasia, LGIEN and high-grade intraepi- need for and the development of CAC in thelial neoplasia, HGIEN) or the need to undergo colec- ulcerative colitis (UC) patients. tomy was also significantly related to disease duration Methods A cohort of n = 434 UC patients was evaluated. (P = 0.006, P = 0.002, respectively). Established anti- Data analysis was performed by univariate and multivariate inflammatory (e.g., 5-ASA, anti-TNF-a) sig- logistic regression. Odds ratios (OR) and 95 % confidence nificantly reduced the risk of both dysplasia and CAC intervals (CI) were calculated, and significance was (P = 0.02). assessed by the likelihood ratio test. Conclusions Despite the use of modern therapies for UC, Results Mean patient age at UC diagnosis was CAC rates remain high. In our study, risk factors included 45.7 ± 15.1 years which manifested mainly as pancolitis disease duration while anti-inflammatory therapies reduced (47 %) or left-sided colitis (45.2 %). CAC was detected in the risk. Effective control of the intestinal inflammation ten patients (2.3 %). UC disease duration was strongly also reduced the disease burden as indicated by decreased risk of requiring colectomy, underscoring the need for sufficient surveillance and anti-inflammatory therapies. Tobias M. Nowacki, Markus Bru¨ckner, Matthias Ross and Dominik Bettenworth have contributed equally to this work.

T. M. Nowacki Á M. Bru¨ckner Á P. Tepasse Á F. Pott Á M. Eveslage K. Hengst Á D. Bettenworth (&) Institute of Biostatistics and Clinical Research, University Department of Medicine B, University Hospital Mu¨nster, Mu¨nster, Mu¨nster, Germany Albert-Schweitzer-Campus 1, 48149 Mu¨nster, Germany e-mail: [email protected] e-mail: [email protected] N. H. Thoennissen T. M. Nowacki Department of Hematology and Oncology, Hubertus Wald e-mail: [email protected] Tumorzentrum, University Cancer Center Hamburg, University M. Bru¨ckner Medical Center Hamburg-Eppendorf, Hamburg, Germany e-mail: [email protected] e-mail: [email protected] P. Tepasse M. Ross e-mail: [email protected] Christophorus-Kliniken GmbH, St.-Vincenz-Hospital Coesfeld, F. Pott Coesfeld, Germany e-mail: [email protected] e-mail: [email protected] K. Hengst e-mail: [email protected]

123 Dig Dis Sci (2015) 60:492–501 493

Keywords Inflammatory bowel disease (IBD) Á delayed or missed in 18–27 % of UC patients [19, 20], a Ulcerative colitis (UC) Á Colitis-associated cancer (CAC) Á problem compounded by low patient uptake of surveillance Á Surveillance programs offered to those with long-standing extensive colitis [21]. Repetitive colonic inflammation not only fuels the Introduction disease burden for the individual patient but also increases the risk to require colectomy and colectomy- Chronic colonic inflammation is a well-known risk factor free survival rates have been proposed as a measurement for the development of colorectal malignancy [1]. Patients of therapeutic effectiveness [22]. Yet, risk factors for the with ulcerative colitis (UC) are at particularly increased need for colectomy are not well established. risk of developing colitis-associated cancer (CAC) [2, 3]. The aim of the present study was to evaluate risk factors While the incidence of sporadic colorectal cancer (CRC) for colectomy, development of dysplasia (LGIEN, HGIEN) is increased from the fifth decade of life, and maximal in and colon cancer, as well as to determine protective mea- the eighth decade, the median age of CAC diagnosis is sures in a relatively large cohort of UC patients at a tertiary 43 years [4, 5]. CAC has a significant impact on UC referral center. patients, with cause-specific mortality due to CRC sig- nificantly higher in UC patients compared to the back- ground population. Munkholm et al. [6] demonstrated Methods 15 % of all in UC patients to be a consequence of CRC. Validated risk factors for CAC are helpful to Enrollment of Patients and Data Collection stratify the individual patients’ risk and include severity and duration of disease, primary sclerosing cholangitis We retrospectively analyzed 434 patients treated as (PSC) as well as a positive family history for CRC [7– inpatients or outpatients between 2002 and 2013 at the 11]. Department of , University Hospital of Historically, the risk of CAC was first reported in a Mu¨nster, a tertiary referral center for IBD patients in pediatric cohort. Devrode et al. [12] found the risk to be northwestern Germany. Patients with UC as principle 20 % per decade beginning 10 years after diagnosis. More diagnosis were identified by searching for code K51 recently, a large meta-analysis including 116 studies (International Classification of Diseases, version 10, revealed incidence rates corresponding to cumulative ICD-10). Additionally, patients with both UC and colo- probabilities of 2 % by 10 years, 8 % by 20 years and rectal cancer were identified by search for codes C18; 18 % by 30 years of disease duration [8]. In the same C19; and C20. Diagnosis of UC was based on clinical study, the overall prevalence of CRC was reported to be presentation, exclusion of infectious and colitis 3.7 % (95 % CI 3.2–4.2 %). In contrast, a recent study as well as endoscopic and histopathological findings. from Denmark found no increased incidence of CAC and Data on demographics (age and sex), disease character- was confirmed by another study including 32,991 UC istics including duration (time elapsed since initial patients that found no increase in CAC over a 30-year diagnosis), disease severity (assessed by Mayo score) and period [13, 14]. In accordance with these findings, a more manifestations (location and extent of disease (deter- recent meta-analysis by Jess et al. [15] only found CAC mined as the maximal extent of colonic involvement: occurring in 1.6 % of UC patients within 14 years of , left-sided colitis or pancolitis according to the diagnosis which is significantly less compared to the Montreal classification [23])) were collected from studies described earlier [8]. patient’s clinical records (shown in Tables 1, 2)ina Endoscopic surveillance programs for detection of pre- cross-sectional analysis. Additionally, data on colitis- cancerous dysplasia have been evaluated for IBD patients related proctocolectomy, presence and grade of any and been proven to facilitate a reduced colorectal carci- dysplasia and CAC, as well as colitis-related specific noma-related mortality rate [16, 17]. Current guidelines anti-inflammatory and immunosuppressive medication recommend to begin screening , including were collected. random every 10 cm of bowel, in UC patients with Of 455 patients, 21 patients (5.2 %) were excluded from more than one-third of the colon involved 8–10 years after our analysis due to insufficient data being available; leav- initial diagnosis [18]. Subsequently, colonoscopy is sug- ing a total 434 patients included. Of these patients, any gested every 1–2 years in UC patients, and annually in with missing or unavailable data within a single category patients diagnosed with PSC. However, even when led to exclusion of that patient from that specific analysis screening complies with guidelines, CAC diagnosis is only, rather than from the study.

123 494 Dig Dis Sci (2015) 60:492–501

Table 1 Demographic and clinical data of patients with UC as Table 2 Demographic and clinical data of patients with CAC principle diagnosis (n = 434) (n = 10) Mean age ± SD 45.7 ± 15.1 Mean age ± SD 54.7 ± 17.3 Males/females 255 (58.8)/179 Males/females 5/5 (41.2) Max. extent of disease Max. extent of disease Proctitis 0 Proctitis 27 (7.8) Proctosigmoiditis/colitis of left colon 4 Proctosigmoiditis/colitis of left colon 156 (45.2) Pancolitis 6 Pancolitis 162 (47) Duration of disease Duration of disease Initial diagnosis 0 Initial diagnosis 23 (6) 1–8 years 1 1–8 years 189 (49.3) 9–15 years 1 9–15 years 84 (21.9) [15 years 8 [15 years 87 (22.7) Mayo score Mayo score 0–3 4 0–3 136 (44.2) 4–6 1 4–6 51 (16.6) 7–9 – 7–9 61 (19.8) 10–12 2 10–12 60 (19.5) Unknown 3 Presentation of dysplasia Presentation of dysplasia No dysplasia detected 322 (92.3) No dysplasia detected 0 Low-grade intraepithelial neoplasia 12 (3.4) Low-grade intraepithelial neoplasia (LGIEN) 0 (LGIEN) High-grade intraepithelial neoplasia (HGIEN) 0 High-grade intraepithelial neoplasia 5 (1.4) Invasive colon 10 (HGIEN) Proctocolectomy Invasive colon carcinoma 10 (2.3) Yes 10 Colitis-associated colectomy No 0 Yes 74 (17.4) Extraintestinal manifestation No 352 (82.6) No extraintestinal manifestation 6 Extraintestinal manifestation Arthropathy 1 No extraintestinal manifestation 322 (80.5) nodosum 0 Arthropathy 40 (10) Primary sclerosing cholangitis 3 5 (1.3) 0 Primary sclerosing cholangitis 31 (7.8) Uveitis 2 (0.5) Yes 2 Values are given as total numbers with valid percentages in paren- No 8 theses unless stated otherwise Values are given as total numbers unless stated otherwise Assessment of Disease Activity and Histological Analysis from the visit at detection of any dysplasia, respectively. Severity of disease was assessed clinically and endoscop- The extend of disease was determined as the maximal ically using the Mayo scoring system as described else- endoscopically recorded disease extend. Histological ana- where [24]. Briefly, Mayo scoring was performed using the lysis of specimens was performed routinely by following criteria: stool frequency, rectal bleeding, endo- pathologists as per the IBD Dysplasia Morphology Study scopic findings and physician’s global assessment. Data Group [25]. In cases of invasive carcinoma, biopsies were from colonoscopic surveillance were available for 380 sent to a second expert pathologist to confirm diagnosis. patients (87.6 %). For patients without current Specimens were classified as either negative for dysplasia, results, a partial Mayo score was assessed as follows: stool low-grade intra-epithelial neoplasia (LGIEN), high-grade frequency, rectal bleeding and physician’s global assess- intraepithelial neoplasia (HGIEN) or CAC. The histologi- ment. Clinical severity of UC was assessed from the most cal degree of inflammation was not routinely outlined in all recent available patient visit (cross-sectional analysis) or pathologic reports.

123 Dig Dis Sci (2015) 60:492–501 495

Statistical Analysis plus (9.8 %). In total, 56.6 % of UC patients received 5-aminosalicylic acid (5-ASA) med- Mean, standard deviation (SD), range and proportions were ication either alone or in combination with other anti- calculated to analyze patients’ baseline characteristics. The inflammatory drugs. A total of 14.7 % of patients did not probability of CAC, dysplasia and colectomy was modeled receive any anti-inflammatory medication. Between 2002 using univariable binary, univariable multinomial and and 2013, eight patients (1.8 %) were newly diagnosed multivariate logistic regression to assess different severities with CAC, while two additional patients had been diag- of neoplasia. Odds ratios (OR), corresponding 95 % con- nosed prior to 2002. Mean age was 54.7 ± 17.3 years, and fidence intervals (95 % CI), likelihood ratio-type P values the male-to-female ratio was 1:1. Eight patients with CAC for the binary logistic regression and for the multinomial suffered from UC longer than 15 years, one patient logistic regression were calculated. In a second step, mul- between 9 and 15 years and one patient between 1 and tivariate logistic regression analysis with stepwise forward 9 years. The disease extent was as follows: six patients selection of variables was performed; parameters with showed a pancolitis and four patients a left-sided colitis. P values B 0.05 in the likelihood ratio test were included After histological confirmation of invasive carcinoma by a in the model. Inferential statistics are intended to be second pathologist, all patients underwent proctocolec- exploratory, not confirmatory, and were interpreted tomy. Subsequently, four patients received adjuvant che- accordingly. The comparison-wise type-I error rate is motherapy; one patient was treated with supportive care controlled instead of the experiment-wise error rate. The due to poor general condition; and in five patients, further significance level was set at P \ 0.05. No adjustment for treatment was not indicated according to current treatment multiple testing was performed. Statistical analyses were guidelines. Two patient deaths occurred, one due to cho- performed using IBM SPSSÒ Statistics 21 for Windows langiosepsis secondary to biliary obstruction resulting from (IBM Corporation, Somers, NY, USA) and R 2.15.3 (R a coincident Klatskin tumor. No other malignancies were Development Core Team, 2011; R: A language and envi- observed in the patients with CAC. Further clinical char- ronment for statistical computing. R Foundation for Sta- acteristics of patients with CAC are shown in Table 2. tistical Computing, Vienna, Austria). Need for Colitis-Related Colectomy

Results In our cohort of 434 patients, 74 patients (17.4 %) under- went colectomy. These patients were generally older (mean Patients’ Characteristics age 49.4 ± 14.6 vs. 44.7 ± 15.2 years) and had a longer disease duration and more extensive colonic involvement Mean age at presentation was 45.7 ± 15.1 years and ran- compared to non-colectomized patients (Fig. 1). No cor- ged from 20 to 97 years. The male-to-female ratio was relation could be detected for patient gender. Patients with 1.4:1 with 58.8 % male patients. Demographic data are disease duration of more than 15 years had a significantly summarized in Table 1. A total of 49.3 % had a disease greater risk to undergo colectomy compared to patients in duration of 1–8 years; 21.9 % of the patients had UC the first 8 years of the disease (OR 2.4; 95 % CI 1.3–4.2; diagnosed for 9–15 years, and 22.7 % longer than P = 0.005). According to the Montreal classification, the 15 years. With regard to disease extension, only 7.8 % of extent of disease was clearly associated with the need for patients were diagnosed with ulcerative proctitis, whereas colectomy in our cohort. The risk of colectomy was highest 45.2 % were diagnosed with left-sided colitis and 47 % in patients with pancolitis (OR 2.9; 95 % CI 1.6–5.5; with pancolitis. A complete Mayo score could be assessed P = 0.001) while anti-inflammatory or immunosuppres- in 308 (71 %) patients, and 44.2 % of these patients scored sive treatment significantly reduced the risk of colectomy between 0 and 3, reflecting mild disease activity, 16.6 % (OR 0.2; 95 % CI 0.1–0.4; P \ 0.001). Particularly, scored between 4 and 6, while 19.8 % scored 7–9, and patients treated with mesalazine had a markedly lower risk 19.5 % scored 10–12 points. Colectomy was performed in of requiring colectomy as compared to those without 17.4 % of patients due to refractory disease or detected therapy (OR 0.2; 95 % CI 0.1–0.4; P \ 0.001). This was colonic dysplasia, while 352 patients (82.6 %) were treated also true when adjusted for disease severity reflected by conservatively. In addition, 78 patients (19.5 %) showed partial Mayo score. In the multivariate logistic regression additional extraintestinal manifestations of UC including analysis, treatment with 5-ASA was found to be an inde- (10 %) and PSC (7.8 %). pendent protective factor. Furthermore, the risk to require The most frequently administered anti-inflammatory colectomy increased with an elevated partial Mayo score and/or immunosuppressive medication was mesalazine (data not shown). Medical treatment with thiopurines also (18.3 %), mesalazine with oral (13.3 %) or reduced the risk of requiring colectomy (OR 0.3; 95 % CI 123 496 Dig Dis Sci (2015) 60:492–501

Fig. 2 The risk to require colectomy is increased in more extensive and greater duration of disease, while anti-inflammatory treatment is protective. Multinomial logistic regression was applied to calculate odds ratios (with 95 % CI) for several risk factors for patients to Fig. 1 Frequency of colectomy is highest in patients with pancolitis. require colectomy. While anti-inflammatory treatment proved to be Data analyzed by stepwise forward logistic regression for colectom- protective, a longer disease duration (more than 15 years vs. the first ized and non-colectomized patients are depicted with regard to extent 8 years of disease) and more extensive disease (pancolitis vs. of disease (left-sided colitis and pancolitis) and immunosuppressive proctitis) increased the risk significantly therapy. Shown are percentages of patients applicable for each category

0.1–0.8; P = 0.01), whereas treatment with anti-TNF antibodies did not show significant effects (OR 0.8; 95 % CI 0.2–2.7; P = 0.659). Additionally, we employed multivariate logistic regres- sion analysis with stepwise forward selection of variables to identify protective as well as risk factors for colectomy. In support of our previous observations, this model iden- tified greater disease severity (partial Mayo score OR 1.2; 95 % CI 1.1–1.4; P = 0.001) as well as more extensive disease (left-sided colitis OR 3.3; 95 % CI 0.4–28.4; pancolitis OR 9.3; 95 % CI 1.1–79.0; P = 0.003) and a longer duration of disease (duration [ 15 years OR 2.1; 95 % CI 0.9–5.0; P = 0.032) to increase the risk of colectomy. In contrast, any pharmaceutical therapy (OR Fig. 3 The frequency of dysplasia development increases with longer 0.1; 95 % CI 0.0–0.3; P \ 0.01) or a shorter duration of disease duration. Dysplasia and invasive carcinoma were detected in disease (duration 9–15 years OR 0.4; 95 % CI 0.1–1.4) 27 out of 349 (7.7 %) patients during screening colonoscopy. Shown decreased the risk of colectomy (Fig. 2). are percentages of patients with dysplastic lesions (LGIEN and/or HGIEN) stratified according to disease duration, indicating a signif- icant association of dysplasia development and disease duration Prevalence of Dysplasia (LGIEN/HGIEN)

Dysplasia (LGIEN and/or HGIEN) was detected in 17 patients, while invasive carcinoma was detected in 10 out (P = 0.006). In accordance with the binary logistic of 349 patients during screening colonoscopy. A total of regression analysis, multinomial regression revealed a 48.1 % of affected patients (13 of 27) had a disease dura- significant association of disease duration with the devel- tion of more than 15 years (Fig. 3), and binary logistic opment of dysplasia and consecutive carcinoma. In com- regression revealed that disease duration was significantly parison with patients with CAC, all other UC patients had a associated with an increased risk to develop dysplasia. 23.8 times higher chance of remaining without dysplasia in There was an increased chance to develop dysplasia nine the first 8 years after diagnosis (OR 23.8; 95 % CI years after diagnosis (OR 1.6; 95 % CI 0.5–4.8) and after 2.9–193.6; P \ 0.01) and a 8.5 times higher chance after more than 15 years (OR 4.3; 95 % CI 1.8–10.5) as com- 9–15 years of disease (OR 8.5; 95 % CI 1–69.9; P \ 0.01). pared to patients with a disease duration of \8 years Furthermore, the risk of UC patients without CAC to

123 Dig Dis Sci (2015) 60:492–501 497

‘‘merely’’ develop LGIEN was tenfold higher in the first 8 years (OR 10.0; 95 % CI 0.9–117.0) and sixfold higher after 9–15 years of disease (OR 6.0; 95 % CI 0.5–77.8). There was no significant relationship between the development of dysplasia and the extent of disease (P = 0.405). However, a clear tendency toward an increased number of dysplastic lesions was documented in case of extended colonic involvement compared to patients suffering from proctitis: left-sided colitis (OR 2.0) versus pancolitis (OR 2.9). The risk to develop dysplasia signifi- cantly increased after 10 years of disease (OR 1.4; 95 % CI 1.1–1.8; P = 0.004). In contrast, sex or partial Mayo score Fig. 4 The risk to develop dysplasia is increased by longer disease did not show an association with the development of duration, while anti-inflammatory treatment is protective. Stepwise dysplasia. forward logistic regression was applied to calculate odds ratios (with As chemoprotective effects of therapeutic drugs are 95 % CI) for several risk factors for patients to develop dysplasia. reported [26], the impact of anti-inflammatory and/or While anti-inflammatory treatment proved to be protective, duration of disease (duration 9–15 years OR 1.6; 95 % CI 0.5–5.4; dura- immunosuppressive medication on CAC development was tion [ 15 years OR 2.7; 95 % CI 0.9–7.8; P = 0.19) and severity of assessed. Of note, anti-inflammatory/immunosuppressive disease (Partial Mayo score OR 1.04; 95 % CI 0.9–1.2; P = 0.555) therapy significantly reduced the risk of dysplasia devel- were identified as independent risk factors for the development of opment (OR 0.4; 95 % CI 0.2–0.8; P = 0.022). These dysplasia results were confirmed by multinomial regression analysis showing a 4.2 times higher chance for patients without medical therapy to develop dysplasia or invasive carci- increased as compared to the first 8 years of disease (OR noma (OR 4.2; 95 % CI 1.2–15.6). Finally, in a process of 2.5; 95 % CI 0.2–41.1). In contrast, disease duration of variable selection, a multivariable model identified the more than 15 years highly increased the chance by a factor variables ‘‘age (per 10 years)’’ (OR 1.3; 95 % CI 1.0–1.8; 21 (OR 21.4; 95 % CI 2.6–173.6; P \ 0.001). In accor- P = 0.092) and ‘‘anti-inflammatory and/or immunosup- dance with our observation regarding dysplasia develop- pressive therapy’’ (OR 0.3; 95 % CI 0.1–0.8; P = 0.026), ment, no significant relationship was found between the duration of disease (duration 9–15 years OR 1.6; 95 % CI development of CAC and the extent of disease (P = 0.37, 0.5–5.4; duration [ 15 years OR 2.7; 95 % CI 0.9–7.8; Fig. 5). Anti-inflammatory and/or immunosuppressive P = 0.19) and severity of disease (Partial Mayo score OR treatment significantly reduced the chance to develop CAC 1.04; 95 % CI 0.9–1.2; P = 0.555) to explain the devel- (OR 0.3; 95 % CI 0.1–1.0), whereas no significant effect opment of dysplasia best (Fig. 4). could be observed by a single medication like 5-ASA (OR 0.9; 95 % CI 0.2–3.6; P = 0.912), thiopurines (OR 0.7; Carcinoma 95 % CI 0.1–3.3; P = 0.603) or anti-TNF (OR 1.6; 95 % CI 0.2–13.8; P = 0.675). In the univariate analysis, no Eight patients (1.8 %) were newly diagnosed with CAC significant relation was found between development of between 2002 and 2013, while two additional patients of CAC and the severity of disease as reflected by partial the analyzed cohort had been diagnosed with CAC prior to Mayo score (P = 0.82), as well as gender (P = 0.57). The 2002. Histological analysis proved all tumors were ade- variable ‘‘age’’ tended to have an influence (P = 0.07) with nocarcinomas. Three patients with CAC showed PSC as an OR of 1.04 (95 % CI 1.0–1.1) meaning that a difference extraintestinal manifestation of UC, underscoring its role as of 10 years of age increased the chance for carcinoma a risk factor for carcinoma development in this study (OR about 42 %. Interestingly, patients receiving medical 5.6; 95 % CI 1.3–23.7). On the other hand, logistic therapy including anti-TNF antibodies tend to have a regression did not reveal extraintestinal manifestation in slightly increased risk for carcinoma development, general to be a significant risk factor for the development although not reaching statistical significance (OR 1.6; of CAC (P = 0.439). At the time of cancer diagnosis, 95 % CI 0.2–13.8; P = 0.69) and probably being con- tumor manifestation was categorized as UICC (Union founded by the factor severity of disease represented by internationale contre le cancer) stage IIa in five patients, Partial Mayo score: Patients with mesalazine as single stage IIIa in two patients and stage IV in one patient. therapy were categorized with an average score of one, Duration of disease significantly increased the risk to whereas patients treated with anti-TNF antibodies had an develop colon carcinoma. In detail, the chance of CAC average score of four points. Finally, a variable selection development after 9–15 years of disease was significantly generated a final multivariable model choosing the 123 498 Dig Dis Sci (2015) 60:492–501

be strongly associated with the highest risk of cancer development in our single-center experience. In addition, our data show that anti-inflammatory medication use can decrease the risk of CAC development. Existing epide- miological data on the incidence rate of CAC in UC patients are conflicting. While historic data hint at markedly increased rates (3.7 %), other studies found comparable risk rates or only slightly increased rates of 1.6 or 1.7 % [8, 15, 27]. Although direct comparisons of crude colorectal cancer occurrence rates and incidence ratios in our study and population-based studies are not possible, the observed occurrence rate of 1.8 % tends toward the more recent reported incidence rates [15]. Despite the limitations of this study in determining inci- Fig. 5 Development of CAC occurs mainly in patients with more dence rates, the observed occurrence rates for CAC were extensive disease. Ten patients out of 434 were diagnosed with CAC. still lower than initially reported incidence ratios [8] Depicted is the total number of patients for the respective extent of although our hospital is a tertiary referral center for IBD disease specified on the x-axis patients. This could be explained by either methodologi- cal aspects (retrospective hospital-based vs. population- based studies, heterogeneous follow-up periods) or truly decreased risk factors due to improved (therapeutic) dis- ease control and surveillance with increased rates of colectomy. Approximately 20 % of UC patients will require surgi- cal treatment and interestingly, and recent studies docu- ment that the introduction of immunosuppressive therapy did not change colectomy rates over time [28]. This finding is in line with the colectomy rates of *17 % in our cohort. We found a significant association between severity and duration of disease and the risk to require colectomy. Additionally, effective control of the colonic inflammation was demonstrated to be a protective effect both, on the risk Fig. 6 The risk to develop CAC is increased by longer disease to develop CAC, dysplasia and to require colectomy. One duration, while anti-inflammatory treatment is protective. Stepwise forward logistic regression was applied to calculate odds ratios (with might postulate that indeed more effective anti-inflamma- 95 % CI) for several risk factors for patients to develop CAC. While tory therapies and the consistently high rates of colectomy anti-inflammatory treatment proved to be protective, duration of contribute to the recently observed reduced risk to develop disease was identified as an independent risk factor for the develop- CAC. Considering colectomy-free survival as one aspect of ment of CAC intestinal damage, our study identified several risk factors including severity, extent and duration of disease, while anti-inflammatory therapy significantly reduced the risk. variables ‘‘anti-inflammatory and/or immunosuppressive Accordingly, these aspects of disease behavior may be used therapy’’ (OR 0.4; 95 % CI 0.1–1.5; P = 0.170), ‘‘duration for individual risk stratification and to detect severe courses of disease 9–15 years’’ (OR 2.5; 95 % CI 0.2–40.1) and of disease and additionally underscore the need for tight ‘‘duration of disease of more than 15 years’’ (OR 18.8; control of the colonic inflammation. 95 % CI 2.3–155.0; P = 0.001) to explain the develop- We found a strong association between the duration of ment of CAC best (Fig. 6). UC and the prevalence of CAC. As compared to the risk of cancer development in the first 8 years after diagnosis, the risk significantly increased between 9 and 15 years and was Discussion even higher after 15 years of disease. Although a British cohort surveillance report found no increased carcinoma In this study, we documented ten newly diagnosed cases rates with increasing disease duration [29], this observation of CAC in our cohort of 434 UC patients. In accordance confirms previously reported disease duration as a main with established risk factors, we found duration of UC to risk factor for colorectal malignancy [8]. 123 Dig Dis Sci (2015) 60:492–501 499

In the light of recent molecular insights into the tumor CAC development in patients with anti-inflammatory and/ development, the relation between long-lasting colonic or immunosuppressive treatment. Of note, one study inflammation and the genesis of CAC seems particularly reported a higher CAC risk in UC patients with extended reasonable. Several studies demonstrate the link between duration of 5-ASA medication [36]. Moreover, in our the development of CAC in UC with chronic inflammation study, patients receiving anti-TNF treatment tended to have inducing genomic alterations, e.g., through executors of a slightly increased risk of carcinoma development. How- oxidative stress such as reactive oxygen and nitrogen ever, one might suggest that patients with continuous need species causing microsatellite instability and impairment of of 5-ASA medication or anti-TNF treatment due to a DNA repair processes [30, 31]. However, more recently, chronic actively course of disease with ongoing inflam- direct effects of the inflammation itself on tumor devel- mation are at increased risk of CAC due to the compara- opment were revealed. For example, the pro-inflammatory tively high disease activity. Confirmatively, no increase in interleukin-6 (IL-6) has an impact on the patho- CAC in anti-TNF-treated patients has been found yet [37]. genesis of CAC. It was shown that UC patients with active In conclusion, 5-ASA treatment in UC patients without flare of disease had significantly more IL-6-positive intes- active disease, only for reasons of chemoprevention could tinal epithelial cells as compared to those patients with not be recommended in general [38]. Since inflammation is quiescent disease. Functionally, it may be hypothesized a crucial factor for CAC development, early aggressive that long-term inflammatory changes in these patients anti-inflammatory therapy is recommended and may multiply inflammation-induced premalignant alterations include long-term 5-ASA treatment, as well as early anti- ultimately leading to the development of CAC. TNF treatment or immunosuppressive therapy with Interestingly, in our patient cohort, we did not find a azathioprine. significant correlation between the development of dys- However, even if UC patients are taking a chemopre- plasia and/or CAC and the extent of disease or severity of ventive medication, there is still a considerable amount of disease. These findings may have been influenced by dif- patients developing CAC. Therefore, endoscopic dysplasia ferent definitions of disease extension (either endoscopic or surveillance programs have been developed proposing histological), the cross-sectional nature of disease severity surveillance in extensive colitis beginning at 8–10 years analysis and by the fact that IBD-related and sporadic after onset of disease and in case of left-sided colitis after neoplasia were cumulatively evaluated in the present ana- 15 years, respectively [39–41]. Although accordant sur- lysis without histological classification of the degree of veillance strategies have been implemented in our clinic, inflammation. However, the association of inflammatory data on the stringency of adherence and quality of endo- severity and the risk to develop dysplasia is well estab- scopic procedures could not be accessed for all patients, lished [32], and in our study, there was a clear tendency of since surveillance were also carried out in increased risk of dysplasia in case of extended colonic referring medical centers according to patient preferences. involvement as compared to proctitis: left-sided colitis (OR Despite the mentioned limitations of this study in corre- 2.0) and pancolitis (OR 2.9). Vice versa, proctitis and lating extend of disease with development of dysplasia or proctosigmoiditis did not increase the risk of CAC [33], CAC, our data suggest that a strict classification of and confirmatively, none of our carcinoma patients suf- pancolitis versus left-sided colitis may not be appropriate fered from proctitis or proctosigmoiditis. In addition, PSC in all cases. As a consequence, patients with left-sided as another established risk factor for CAC was present in colitis and high disease activity may also be screened three of ten (30 %) patients, but not reaching a significant 8 years after onset of disease. In addition, repetitive en- association. Nevertheless, taken into account that data from doscopies are not only used for cancer surveillance strat- 434 UC patients were analyzed, we hypothesize that larger egies but have also gained importance with the advent of cohorts might confirm the role of extensive disease and endoscopic endpoints in the evaluation of treatment effi- PSC as independent risk factors. cacy and determination of remission in terms of mucosal Due to the abundant evidence in the literature proving healing as a predictor for fewer clinical relapse and lower that chronic inflammation drives carcinogenesis, there is a colectomy rates [42]. pressing need for effective anti-inflammatory drugs not Due to the fact that our data are based on a single-center only to relieve the patients’ symptoms but also to prevent study, the general validity may be limited. However, it is CAC. For several years, 5-ASA was supposed to be the the best way to study regional cohorts and to avoid heter- most potent chemopreventive drug available. However, ogeneity of large multicenter studies. Furthermore, the available data for the chemopreventive effect of 5-ASA are observed risk of CAC may be influenced by the individual complex with some studies confirming a protective effect surveillance strategies, since a tight follow-up program while other studies could not reproduce these findings [26, including screening colonoscopy will increase the likeli- 34, 35]. Yet, our study demonstrated a decreased risk of hood to detect CAC. Interestingly, the insightful meta- 123 500 Dig Dis Sci (2015) 60:492–501 analysis by Eaden et al. [8] indicated geographical differ- 5. Eddy DM. Screening for colorectal cancer. Ann Intern Med. ences in the risk of CAC with a low or reduced risk of CAC 1990;113:373–384. 6. Munkholm P. Review article: the incidence and prevalence of in Scandinavia. In detail, the authors hypothesized that this colorectal cancer in inflammatory bowel disease. Alim Pharmacol may be due to true genetic or environmental differences as Ther. 2003;18:1–5. well as to a more aggressive UC treatment approach in 7. Claessen MM, Lutgens MW, van Buuren HR, et al. More right- Scandinavia. In addition, the data from Scandinavia were sided IBD-associated colorectal cancer in patients with primary sclerosing cholangitis. Inflamm Bowel Dis. 2009;15:1331–1336. collected from hospitals treating a defined catchment area 8. Eaden JA, Abrams KR, Mayberry JF. The risk of colorectal so that UC patients with different disease severity are cancer in ulcerative colitis: a meta-analysis. Gut. 2001;48:526– included, unlike in tertiary referral centers mostly treating 535. more severe cases of disease [8]. Comorbidity and co- 9. Rutter M, Saunders B, Wilkinson K, et al. Severity of inflam- mation is a risk factor for colorectal neoplasia in ulcerative medication might also have an impact on CAC frequency colitis. Gastroenterology. 2004;126:451–459. since there are data indicating that aspirin might reduce the 10. Soetikno RM, Lin OS, Heidenreich PA, Young HS, Blackstone risk of CAC [43]. However, the long-term safety still needs MO. Increased risk of colorectal neoplasia in patients with pri- to be confirmed. Interestingly, some studies have found an mary sclerosing cholangitis and ulcerative colitis: a meta-ana- lysis. Gastrointest Endosc. 2002;56:48–54. increased risk of CAC in males [15], yet this effect was 11. Askling J, Dickman PW, Karlen P, et al. Family history as a risk confined to patients with follow-up greater than 10 years factor for colorectal cancer in inflammatory bowel disease. and patients younger than 45 years of age at first diagnosis Gastroenterology. 2001;120:1356–1362. while other studies found no significant correlation [2, 44]. 12. Devroede GJ, Taylor WF, Sauer WG, Jackman RJ, Stickler GB. Cancer risk and life expectancy of children with ulcerative colitis. Although gender-specific differences in malignancy N Engl J Med. 1971;285:17–21. development have been reported in multiple malignancies, 13. Jess T, Simonsen J, Jorgensen KT, Pedersen BV, Nielsen NM, including CRC, our study as well could not detect differ- Frisch M. Decreasing risk of colorectal cancer in patients with ences, indicating the need for further specific population- inflammatory bowel disease over 30 years Gastroenterology. 2012;143:375–381 e371; quiz e313-374. based studies to optimize surveillance strategies possibly 14. Winther KV, Jess T, Langholz E, Munkholm P, Binder V. Long- taking gender-specific risk constellations into account. term risk of cancer in ulcerative colitis: a population-based cohort In this cohort, an increased risk of CAC was observed. study from Copenhagen County. Clin Gastroenterol Hepatol. Due to the fact that uncontrolled inflammation is one major 2004;2:1088–1095. 15. Jess T, Rungoe C, Peyrin-Biroulet L. Risk of colorectal cancer in source of premalignant alterations, the new IBD treatment patients with ulcerative colitis: a meta-analysis of population-based paradigm of early intervention (accelerated step up thera- cohort studies. Clin Gastroenterol Hepatol. 2012;10:639–645. peutic strategy) including immunosuppressive therapy 16. Choi PM, Nugent FW, Schoetz DJ Jr, Silverman ML, Haggitt RC. might eventually reduce the risk of CAC in the long-term. Colonoscopic surveillance reduces mortality from colorectal cancer in ulcerative colitis. Gastroenterology. 1993;105:418–424. The role of a lifelong treatment with chemopreventive 17. Vleggaar FP, Lutgens MW, Claessen MM. Review article: The drugs, as well as appropriate endoscopic surveillance pro- relevance of surveillance endoscopy in long-lasting inflammatory gram, still needs to be further elucidated. bowel disease. Alim Pharmacol Ther. 2007;26:47–52. 18. Farraye FA, Odze RD, Eaden J, et al. AGA medical position Acknowledgments We thank Faekah Gohar for proofreading the statement on the diagnosis and management of colorectal neo- manuscript. D. Bettenworth was supported by a research fellowship plasia in inflammatory bowel disease. Gastroenterology. from the Faculty of Medicine, Westfa¨lische Wilhelms-Universita¨t 2010;138:738–745. Mu¨nster. M. Bru¨ckner was supported by a ‘‘Gerok’’ rotational posi- 19. Jess T, Gamborg M, Munkholm P, Sorensen TI. Overall and tion of the Deutsche Forschungsgemeinschaft (DFG SFB1009B8). cause-specific mortality in ulcerative colitis: meta-analysis of population-based inception cohort studies. Am J Gastroenterol. Conflict of interest There was no financial support for this study, 2007;102:609–617. and the authors declare no conflict of interest. 20. Lutgens MW, Vleggaar FP, Schipper ME, et al. High frequency of early colorectal cancer in inflammatory bowel disease. Gut. 2008;57:1246–1251. 21. Vienne A, Simon T, Cosnes J, et al. Low prevalence of colono- References scopic surveillance of inflammatory bowel disease patients with longstanding extensive colitis: a clinical practice survey nested in 1. Grivennikov SI, Greten FR, Karin M. Immunity, inflammation, the CESAME cohort. Alim Pharmacol Ther. 2011;34:188–195. and cancer. Cell. 2010;140:883–899. 22. Subramaniam K, Richardson A, Dodd J, Platten J, Shadbolt B, 2. Ekbom A, Helmick C, Zack M, Adami HO. Ulcerative colitis and Pavli P. Early predictors of colectomy and long-term mainte- colorectal cancer. A population-based study. N Engl J Med. nance of remission in ulcerative colitis patients treated using anti- 1990;323:1228–1233. TNF therapy Intern Med J 2014. 3. Ekbom A, Helmick C, Zack M, Adami HO. Increased risk of 23. Satsangi J, Silverberg MS, Vermeire S, Colombel JF, Satsangi J. large-bowel cancer in Crohn’s disease with colonic involvement. The Montreal classification of inflammatory bowel disease: Lancet. 1990;336:357–359. controversies, consensus, and implications. Gut. 2006;55: 4. Choi PM, Zelig MP. Similarity of colorectal cancer in Crohn’s 749–753. disease and ulcerative colitis: implications for carcinogenesis and 24. Schroeder KW, Tremaine WJ, Ilstrup DM. Coated oral 5-am- prevention. Gut. 1994;35:950–954. inosalicylic acid therapy for mildly to moderately active

123 Dig Dis Sci (2015) 60:492–501 501

ulcerative colitis. A randomized study. N Engl J Med. 1987;317: study from Copenhagen county, Denmark and Olmsted county, 1625–1629. Minnesota. Am J Gastroenterol. 2007;102:829–836. 25. Riddell RH, Goldman H, Ransohoff DF, et al. Dysplasia in 35. Ullman T, Croog V, Harpaz N et al. Progression to colorectal inflammatory bowel disease: standardized classification with neoplasia in ulcerative colitis: effect of mesalamine. Clin Gas- provisional clinical applications. Hum Pathol. 1983;14:931–968. troenterol Hepatol. 2008;6:1225–1230; quiz 1177. 26. Velayos FS, Terdiman JP, Walsh JM. Effect of 5- 36. Bernstein CN, Nugent Z, Blanchard JF. 5-Aminosalicylate is not use on colorectal cancer and dysplasia risk: a systematic review chemoprophylactic for colorectal cancer in IBD: a population and metaanalysis of observational studies. Am J Gastroenterol. based study. Am J Gastroenterol. 2011;106:731–736. 2005;100:1345–1353. 37. Ali RA, Egan LJ. How to manage the risk of colorectal cancer in 27. Lutgens MW, van Oijen MG, van der Heijden GJ, Vleggaar FP, ulcerative colitis. Curr Drug Targets. 2011;12:1424–1432. Siersema PD, Oldenburg B. Declining risk of colorectal cancer in 38. Herfarth H. The role of chemoprevention of colorectal cancer inflammatory bowel disease: an updated meta-analysis of popu- with 5- in ulcerative colitis. Dig Dis. 2012;30: lation-based cohort studies. Inflamm Bowel Dis. 2013;19:789– 55–59. 799. 39. Biancone L, Michetti P, Travis S, et al. European evidence-based 28. Moore SE, McGrail KM, Peterson S, et al. Infliximab in ulcera- Consensus on the management of ulcerative colitis: special sit- tive colitis: the impact of preoperative treatment on rates of uations. J Crohns Colitis. 2008;2:63–92. colectomy and prescribing practices in the province of British 40. Dignass A, Preiss JC, Aust DE, et al. [Updated German guideline Columbia, Canada. Dis Colon . 2014;57:83–90. on diagnosis and treatment of ulcerative colitis, 2011]. Z Gas- 29. Rutter MD, Saunders BP, Wilkinson KH, et al. Thirty-year ana- troenterol. 2011;49:1276–1341. lysis of a colonoscopic surveillance program for neoplasia in 41. Hoffmann JC, Zeitz M, Bischoff SC, et al. [Diagnosis and therapy ulcerative colitis. Gastroenterology. 2006;130:1030–1038. of ulcerative colitis: results of an evidence based consensus 30. Cravo ML, Albuquerque CM, Salazar de Sousa L, et al. Micro- conference by the German society of digestive and metabolic satellite instability in non-neoplastic mucosa of patients with diseases and the competence network on inflammatory bowel ulcerative colitis: effect of folate supplementation. Am J Gas- disease]. Z Gastroenterol. 2004;42:979–983. troenterol. 1998;93:2060–2064. 42. Neurath MF, Travis SP. Mucosal healing in inflammatory bowel 31. Xavier RJ, Podolsky DK. Unravelling the pathogenesis of diseases: a systematic review. Gut. 2012;61:1619–1635. inflammatory bowel disease. Nature. 2007;448:427–434. 43. Velayos FS, Loftus EV Jr, Jess T, et al. Predictive and protective 32. Rubin DT, Huo D, Kinnucan JA et al. Inflammation is an inde- factors associated with colorectal cancer in ulcerative colitis: a pendent risk factor for colonic neoplasia in patients with - case-control study. Gastroenterology. 2006;130:1941–1949. ative colitis: a case-control study Clin Gastroenterol Hepatol. 44. Soderlund S, Granath F, Brostrom O, et al. Inflammatory bowel 2013;11:1601–1608 e1601–1604. disease confers a lower risk of colorectal cancer to females than 33. Lukas M. Inflammatory bowel disease as a risk factor for colo- to males. Gastroenterology. 2010;138:1697–1703. rectal cancer. Dig Dis. 2010;28:619–624. 34. Jess T, Loftus EV, Velayos FS Jr, et al. Risk factors for colorectal neoplasia in inflammatory bowel disease: a nested case-control

123